Pixinsight Workflow

[Niall Saunders]

Is it possible for DBE to 'clip' an image? I did a DBE on M33 and the histogram seems to be lopped off. Granted I did not do due diligence in my background modeling ... most of the frame is galaxy and points that are arguably not background were likely included in the model.

This is the reason I always advise folk to 'examine' the 'background image' that has been subtracted as a result of the DBE process - and it will take an aggressive STF or HISTO stretch to bring out the details of what has been 'removed'. Providing that you feel comfortable with what has been removed then, yes, although the image has most likely been clipped - the pixels that have been clipped are not identifiable as 'structure' in the original image.

Remember, DBE is always under the user's control, even if the user may not actually BE in control So, if you don't like what was removed, it's probably time for a CTRL-Z on the original image - but 'save' the current DBE process as a ProcessIcon, just in case you change your mind. Then RESET the DBE process, completely, and run it again and see if things work better. Try it a few different ways, and settle on the one 'you like the most'.

And, always bear in mind that LESS can sometimes be MORE - so don't always be tempted to overdo DBE

[Luigi]

I should have made sure the background model was correct. My only excuse is that I was excited as these were the first images from the new camera 

I shall go back and rein in all those 'spurious' points. Thanks.

[RBA]

Then RESET the DBE process, completely, and run it again and see if things work better. Try it a few different ways, and settle on the one 'you like the most'.

My suggestion - which isn't better nor worst than Niall's, just different - would be not to reset the parameters. By adjusting the parameters, adding/removing samples, etc. you can influence the resulting background model gradually until you get what you believe you need.

A start over to me is more like "ok I'm way off, let's have a clean start" (which does happen) but then after the start over, I would again go on readjusting things if the new parameters/samples still don't cut the mustard.

Anyway, just sharing experiences. By no means I mean that Niall's suggestion is a bad one, but that in my experience, such reset isn't always necessary. And as always, your mileage may vary 

[Niall Saunders]

Agreed with Rogelio - it isn't always a 'Reset' - sometimes it is more of a 're-think' when it comes to re-applying a process

[Nigel Ball]

My suggestion - which isn't better nor worst than Niall's, just different - would be not to reset the parameters. By adjusting the parameters, adding/removing samples, etc. you can influence the resulting background model gradually until you get what you believe you need.

This is essentially the process I have 'homed' in on.

My DBE is an iterative process.

• Do an initial run of DBE
• examine the Background at Auto STF
• What does the Histogram look like? Is it okay?
• If not then add/remove samples and re run DBE

As I said it is an iterative process